Instructions to use majentik/gemma-4-31B-it-RotorQuant-MLX-2bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use majentik/gemma-4-31B-it-RotorQuant-MLX-2bit with MLX:

# Make sure mlx-vlm is installed
# pip install --upgrade mlx-vlm

from mlx_vlm import load, generate
from mlx_vlm.prompt_utils import apply_chat_template
from mlx_vlm.utils import load_config

# Load the model
model, processor = load("majentik/gemma-4-31B-it-RotorQuant-MLX-2bit")
config = load_config("majentik/gemma-4-31B-it-RotorQuant-MLX-2bit")

# Prepare input
image = ["http://images.cocodataset.org/val2017/000000039769.jpg"]
prompt = "Describe this image."

# Apply chat template
formatted_prompt = apply_chat_template(
    processor, config, prompt, num_images=1
)

# Generate output
output = generate(model, processor, formatted_prompt, image)
print(output)

Notebooks
Google Colab
Kaggle
Local Apps Settings
LM Studio

How to use majentik/gemma-4-31B-it-RotorQuant-MLX-2bit with Pi:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "majentik/gemma-4-31B-it-RotorQuant-MLX-2bit"

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "mlx-lm": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "majentik/gemma-4-31B-it-RotorQuant-MLX-2bit"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent

How to use majentik/gemma-4-31B-it-RotorQuant-MLX-2bit with Hermes Agent:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "majentik/gemma-4-31B-it-RotorQuant-MLX-2bit"

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default majentik/gemma-4-31B-it-RotorQuant-MLX-2bit

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use majentik/gemma-4-31B-it-RotorQuant-MLX-2bit with OpenClaw:

Start the MLX server

# Install MLX LM:
uv tool install mlx-lm
# Start a local OpenAI-compatible server:
mlx_lm.server --model "majentik/gemma-4-31B-it-RotorQuant-MLX-2bit"

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "majentik/gemma-4-31B-it-RotorQuant-MLX-2bit" \
  --custom-provider-id mlx-lm \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

gemma-4-31B-it-RotorQuant-MLX-2bit

Commit History

Remove links to retired duplicate brand variants

ee05058
verified

majentik commited on 10 days ago

Fix quantization metadata: top-level bits 2 (matches tensor data)

2861d44
verified

majentik commited on 21 days ago

Card accuracy pass 2: remove unmeasured speed claims, honest brand labels

efbecfb
verified

majentik commited on Jul 20

Card accuracy sweep: honest brand labeling, remove dead links, upstream KV tip

3be5469
verified

majentik commited on Jul 20

Remove links to retired card-only KV-docs repos

18d380f
verified

majentik commited on Jul 20

docs: upstream-first KV-cache guidance (q8_0/q4_0, mainline Hadamard rotation); fork demoted to experimental

a1c1d3a
verified

majentik commited on Jul 14

docs: remove AWQ variants from family table (AWQ lane paused, repos private)

c032db5
verified

majentik commited on Jul 9

fix: family model links majentik/gemma4-* -> majentik/gemma-4-* (reported by @shramee in gemma-4-31B-TurboQuant-MLX-4bit#1)

aff45ce
verified

majentik commited on Jul 7

docs: Tier 2 polish — variant matrix + quant trade-off

886e60d
verified

majentik commited on May 6

chore(card): enrich YAML frontmatter (pipeline_tag, language, library_name, inference)

401ec9d
verified

majentik commited on Apr 17

Add MLX quantized model with KV cache compression

4c89218
verified

majentik commited on Apr 13

initial commit

496b83a
verified

majentik commited on Apr 13

Commit History

Remove links to retired duplicate brand variants ee05058 verified

Fix quantization metadata: top-level bits 2 (matches tensor data) 2861d44 verified

Card accuracy pass 2: remove unmeasured speed claims, honest brand labels efbecfb verified

Card accuracy sweep: honest brand labeling, remove dead links, upstream KV tip 3be5469 verified

Remove links to retired card-only KV-docs repos 18d380f verified

docs: upstream-first KV-cache guidance (q8_0/q4_0, mainline Hadamard rotation); fork demoted to experimental a1c1d3a verified

docs: remove AWQ variants from family table (AWQ lane paused, repos private) c032db5 verified

fix: family model links majentik/gemma4-* -> majentik/gemma-4-* (reported by @shramee in gemma-4-31B-TurboQuant-MLX-4bit#1) aff45ce verified

docs: Tier 2 polish — variant matrix + quant trade-off 886e60d verified

chore(card): enrich YAML frontmatter (pipeline_tag, language, library_name, inference) 401ec9d verified

Add MLX quantized model with KV cache compression 4c89218 verified

initial commit 496b83a verified

Remove links to retired duplicate brand variants

ee05058
verified

Fix quantization metadata: top-level bits 2 (matches tensor data)

2861d44
verified

Card accuracy pass 2: remove unmeasured speed claims, honest brand labels

efbecfb
verified

Card accuracy sweep: honest brand labeling, remove dead links, upstream KV tip

3be5469
verified

Remove links to retired card-only KV-docs repos

18d380f
verified

docs: upstream-first KV-cache guidance (q8_0/q4_0, mainline Hadamard rotation); fork demoted to experimental

a1c1d3a
verified

docs: remove AWQ variants from family table (AWQ lane paused, repos private)

c032db5
verified

fix: family model links majentik/gemma4-* -> majentik/gemma-4-* (reported by @shramee in gemma-4-31B-TurboQuant-MLX-4bit#1)

aff45ce
verified

docs: Tier 2 polish — variant matrix + quant trade-off

886e60d
verified

chore(card): enrich YAML frontmatter (pipeline_tag, language, library_name, inference)

401ec9d
verified

Add MLX quantized model with KV cache compression

4c89218
verified

initial commit

496b83a
verified